ELAC;
The choices seem to be either, the approach was not vertically stable in the late stages,
That is why I included the possibility that, similar to AA1420, the aircraft got high in the very last stages of the approach during the "inside to outside" perceptual transition, in darkness and heavy rain. "Floating" makes little sense as the energy is not there, as you say. If the aircraft is high, depth perception poor and time is compressing rapidly - (another similarity with AA1420, but not, obviously a causal factor in and of itself), that may explain the distance to touchdown. Nor do we know the power settings across the threshold, either but I wonder - will the N1s be low to idle indicating a higher than normal threshold crossing height, or will we see N1s reducing and then, if the autothrust was still engaged, increasing again in response to autothrust commands to maintain the approach target airspeed; - we do not have that information and cannot discount either notion.
galaxy flyer, whippersnapper; I wouldn't look towards training as "being at fault". I think most would agree, as p51guy states that the need for a baulked landing at this stage of the approach is extremely rare and training time is expensive and has to respect the more common occurences as described in law and what is seen in the FOQA data.
That said, there is a FOQA event one can design which monitors a go-around from a low energy state, (thrust levers closed, N1s below nn%, reducing airspeed) and I have seen it trigger and the go-around has been successful. Clearly, such a decision is made very swiftly and not well into the landing regime! Bear in mind that the Transport Canada statements issue directly from an attempted go-around by an RJ at Fredericton and may not apply to all circumstances and/or all types.
Clearly too,
this is not "advice" but speculative commentary. That said, whenever we do (or used to do), live training circuits and bumps for transition courses etc, we are/were doing go-arounds from this very state every time, (spoilers not armed, though) so we aren't exactly in "test pilot territory". But it is not trained as regular ops and isn't on any syllabus I am aware of. Clearly, the risk in such a decision grows steeply with every passing second, especially if one does not know exactly how much runway is remaining. Someone pointed out 'different colored lights' towards the end of the runway - well, that only applies to the alternating white-red then solid red CLLs and this runway had none of that.
Long landings is something airlines must see in their flight data analysis programs. They occur in all circumstances, for all kinds of reasons. Trying for a smooth landing is one possible cause; if the autothrust isn't disconnected and attempts to "maintain the target airspeed" before the pilot flying catches the thrust levers disconnects and closes them is another. Airspeeds much higher than Vref plus the usual additives is more common than one would expect, and tailwind landings are not that uncommon either. Also, one cannot discount the effects of TAS vice CAS - on a 30C day, even at sea level, the TAS can be 5kts higher, with a corresponding higher groundspeed.
p51guy; Re your comment regarding Jamaican authorities, I was reluctant to automatically discount independant comments but you said it directly and it was the reason I asked about the sources of the information we have seen in the media. It seems credible however but so far there is very little to go on and far more questions than answers at this stage.